Determinants of Economic Growth in Post-1994 South Africa

By

Shonisani Tshinakaho Mphinyana1 and Matthew Kofi Ocran2

Abstract: This study investigates the determinants of economic growth in South Africa for
the period 1995-2017. This study uses the Autoregressive Distributed Lag (ARDL). Two
models are estimated. The results of the first model indicate that there is a long-run run
relationship between economic growth and independent variables, while for the second
model there is no long-run relationship. The results of the study suggests that FDI has positive
and significant long-run relationship with economic growth. Moreover, the study finds a
negative relationship between exports and growth. Institutional variables have positive,
although insignificant relationship with economic growth. Portfolio investment has negative,
but insignificant relationship with economic growth.

JEL codes: C32, E10

Keywords: Economic growth, Institutions, ARDL model.

1
University of Mpumalanga, contact email: Shonisani.mphinyana@ump.ac.za
2
University of the Western Cape, Contact email: mocran@uwc.ac.za
1
1. Introduction

For many years, economic growth theorists have attempted to explain what factors determine
the level of economic growth. Although there is a wide variety of growth theories, there is little
consensus as to what factors are more important in promoting or hindering economic growth
(Chirwa & Odhlambo 2016). Earlier theories such as the classical, neoclassical and endogenous
growth theories have attempted to explain how economies can grow. While classical and
neoclassical growth theories put much emphasis on resources, the endogenous growth theories
of Lucas (1988), Romer (1990), Barro (1990), Aghion and Howitt (1992) put emphasis on factors
such as investment in human capital, innovation through research and development as main
drivers of growth. These theories however, have still not fully explain the causes of economic
growth.

The failure of classical, neoclassical and endogenous growth theories to explain why other
countries fail to revive their economies, have put these theories into question (Wanjuu & le Roux,
2017; Ocran, 2019).The new institutional economics of growth has been recently dominating the
literature. The new institutional economics revived Marx and Engels’ (1948) view that institutions
are important for economic growth (Tang, Hu & Li ND: 1). As mentioned by Combarnousand
Rougier (ND:1), the fundamental questions that the new institutional economics is attempting to
answer are: how are current institutions in developed countries formed? How do these
institutions affect economic performance and explain divergence? Why have so many countries
been unable to create rules and norms to promote economic growth and social progress? How
can the survival of inefficient institutions be explained? The new institutional economics
envisages an environment whereby institutions are seen as efficient and stable entities. Those
institutions are a set of norms and rules that promote economic growth through enforcing
property rights, formal contracts and business rules. The new institutional economists’ view is
that, institutions provide incentives to economic participants to invest in certain assets and
acquire certain skills Combarnousand Rougier (ND:1)

Another strand of literature is the coordination failure approach, associated with Rosestein-
Rodan’s (1943) work. The coordination failure approach postulate that, when the markets are
left to function on their own, there tends to be disequilibria, which requires some form of
intervention (Ferris & Gawande 2003; Ocran 2019). Ocran (2019) further states that, for
industrialisation to take place in least developed countries there is a need for external sources
to inject capital to rejuvenate poor economies. Combarnousand Rougier (ND:1), indicates that
to overcome coordination failure, government need to intervene in the markets.

2
For almost a decade, South Africa’s economic trajectory has not been favourable. According to
the World Bank, as at 2017 the main drivers of South Africa’s economy were agriculture and
mining. However, the terms of trade in commodity exporting countries are vulnerable and tend
to be volatile (Deaton, 1999; Shah, Ul haq & Farooq, 2015 ). South Africa’s economy performed
well during the commodity boom era just before the 2008 Global Financial Crisis. Because of
high commodity prices in 2007, the annual growth rate reached a peak of 5.4 per cent. However,
these growth outcomes were short-lived. When the commodity boom came to an end in 2014,
the economy grew by a mere 1.5 per cent per annum. At the beginning of 2018, the economy
entered into a technical recession after two consecutive quarters of negative growth were
recorded (Statistic South Africa 2018:2). In the first quarter of 2019, the economy recorded a
negative growth rate of 3.2 (Statistic South Africa 2019). The economy is still characterised by
high unemployment, with the unemployment rate of 29 per cent in the second quarter of 2019.
In 2016, half of the population was still living in poverty, while income inequality remains high
(Statistics South Africa 2019). The current growth trajectory is not adequate to accommodate the
increasing rate of population. This pattern of slow domestic growth however, is not consistent
with the global economic performance. Fedderke (2018) indicates that South Africa lags behind
as compared to other emerging economies, as well as high and middle income, and this due to
structural factors.

In the midst of these low growth outcomes, South Africa’s political and social institutions continue
to stumble. Uncertainties in the political environment have affected economic growth negatively.
The recent review of the Constitution to make it easy to expropriate land without compensation
has had adverse effects on agricultural investment (Fedderke 2018). Moreover, this proposition
to expropriate land without compensation has raised questions on how property rights will be
affected.

The research questions raised by the discussion above are; what are the determinants of
economic growth in post-1994 South Africa? What are the binding constraints behind low
economic growth rates in South Africa? What role does institutions play in promoting or inhibiting
economic growth?

To answer the aforementioned research questions this study intends to make two contributions
to the body of literature. First, this paper intends to contribute to the body of literature by
identifying the main determinants of economic growth in post-1994 South Africa, also taking into
consideration the effects of institution on economic growth. Second, this study aims to contribute
to the body of literature, by providing insight to policymakers as to which factors to focus on to in
order stimulate growth.

3
 The rest of the paper is organised as follows. Section two provides a brief overview of economic
growth trends in post-1994 South Africa. Section three provides literature review. The research
methodology adopted and description of data are outlined in section four. The main findings and
discussion are provided in section five and conclusion in section six.

2. Overview of Economic Performance in Post-1994 South Africa.

Table 1 below shows the trends in macroeconomic variables since 1994. The performance is
linked to different macroeconomic frameworks that have been introduced since 1994. First, the
Reconstruction Development Programme (RDP) was the ANC’s turnaround strategy. The RDP
aimed to stimulate economic growth through increased government expenditure. As a results of
increased government expenditure during this period, the average growth rate was 3.5 per cent,
however unemployment remained relatively high at 25 percent. On the other hand, population
increased at a significant rate at an average of 2.01 percent.

The Growth, Employment and Redistribution (GEAR) was introduced as a macroeconomic

stabilising strategy. Although the strategy was successful in stabilising the economy,
unemployment remained high at the rate of 26 percent. During the GEAR tenure, economic
growth grew at an average rate of 3.2 percent, a slight decline as compared to the RDP tenure.
Growth rate remained moderate at an average of 3 percent and continued throughout to AsgiSA.
During the AsgiSA period, unemployment was relatively low at 23 per cent. This period was
characterised by high commodity prices, which favoured South Africa’s economy. The end of the
commodity boom however, affected the growth rates significantly, with real GDP decreasing to
an average of 1.35 for the period 2013-2018. Moreover, the recent proposal to amend the
constitution to quicken the land reform process has resulted in policy uncertainties, and this is
manifested in growth outcomes.

Table 1: Trends in Macroeconomic Variables Since 1994.

Macroeconomic Period Average Average Average

framework growth rate unemployment Population
rate Growth
RDP 1994-1996 3.5 25 2.01
GEAR 1996-2005 3.2 26 1.36
AsgiSA 2006-2010 3.14 23 1.30
NDP 2013-2018 1.35 26 1.34

4
Source: World Bank (2017, South African Reserve Bank (2019).

3. Literature Review

Empirical literature indicates that there is no consensus on what factors are important in
determining economic growth in developing and developed countries. Earlier empirical studies
conducted by Dollar (1992), Fischer (1992), investigate the macroeconomic determinants of
growth in developing countries. These studies find comparable results, as they confirm that
capital investment and FDI have positive and significant impact on economic growth. Most and
Vann de Berg (1996) add the rate of population growth when investigating the determinants of
economic growth, and find that population growth has negative and significant impact on
economic growth. Using a panel 100 developed and developing countries, Barro (1999)
investigates factors contributing to economic growth for the period 1960-1995, and finds factors
such as democracy, trade openness have positive and significant impact on economic growth.
Moreover, Barro (1999) finds that government consumption expenditure, fertility rate and inflation
have negative impact on economic growth.

Mbulawa (2015), investigate the long-run determinants of economic growth in Zimbabwe, the
results of the study indicate that trade openness have positive impact on economic growth.
Moreover, the author finds that inflation and FDI have negative impact on economic growth.
Although foreign direct investment is expected to have positive impact on economic growth,
Mbulawa (2015) argues that FDI is beneficial to countries with proper property rights laws. A
study conducted by Mansaray (2017) explores the macroeconomic determinants of economic
growth in post-war Sierra Leone for the period 2002-2013. The author establishes whether there
is a long-run or short-run relationship between growth factors and total output. Using the error
correction model, the variables included in the model are Real GDP as a dependent variable,
and FDI, Gross Capital formation, exchange rate, inflation, trade openness, human capital and
interest rate are explanatory variables. The results of the study indicate that FDI, gross capital
formation have positive and significant impact on economic growth, while inflation has a negative
impact on economic growth.

5
A study conducted by Ulku, (2004), uses panel data for 20 OECD countries to investigate
whether innovation through Research and Development (R&D) has positive and significant
impact on economic growth. The results are consistent with the endogenous growth theory and
indicate a positive and significant impact on total output. These results are also confirmed by a
study conducted by Sokolov-Mladenović, Cvetanović and Mladenović (2016). The study shows
that expenditure on R&D has a positive impact on economic growth. Another study conducted
by Masoud (2013), examines the neoclassical growth theory in a number of developing and
developed countries. The study finds that for developing countries, one of the important factors
contributing to economic growth is improving the quality of labour, rather than the quantity of
labour. To do this, Masoud (2013) advocates improving the quality education and health
services.

A qualitative study conducted by Chirwa & Odhlambo (2016) explores the macroeconomic
determinants of growth in a number of developing and developed countries. The results of the
study indicate that in developing countries the key macroeconomic determinants of economic
growth include foreign aid, foreign direct investment, fiscal policy, investment, trade, human
capital development, demographics, monetary policy, natural resources, reforms and
geographic, regional, political and financial factors. For developed countries, Chirwa &
Odhlambo (2016) find that economic growth is driven by factors including physical capital, fiscal
policy, human capital, trade, demographics, monetary policy and financial and technological
factors.

Bonga-Bonga and Ahiakpor (2015) investigates the determinants of economic growth in Ghana.
The main drivers of economic growth in Ghana are; rate of labour force, population density,
inflation, total agriculture production and current account balance. Mongale and Monkwe (2015)
brings the argument closer and investigates the important factors contributing to economic
growth in South Africa. The study uses the vector error correction model to test for any long-run
relationships and impulse response function to explain the responses to shocks among the
variables. The variables included in the model are real GDP as a dependent variable, with
imports, infrastructure investment and exports as independent variable. The results of the study
indicate that exports, imports and investment in infrastructure have a long-run relationship with
growth, although positive or negative. Although the study may have yielded viable results, the
inclusion of fewer variables may arrive in biased conclusion.

A study conducted by Zouhaier and Kefi (2012) studies the effect of institional factors on
economic growth in a panel of 37 developed and developing countries. Using the data from
1975-2000, the results of the study indicates that civil liberties and political rights do not have a
direct relationship with economic growth for both developed and developing countries. Further,
6
Zouhaier and Kefi (2012) suggest that political institutions will have indirect effect on growth,
through investment and human capital. A study conducted by Lehne, Mo and Plekhanov (2014)
studies how institutions affect economic performance. The results of the study support the
theoretical assertion that institutions matter for economic growth. The results of the study also
indicate that democracy and a country’s history matter for economic growth.

Khemakhem and Abida (2016) investigates the causal relationship between foreign direct
investment, economic freedom and economic growth in a panel of four North African countries.
Using Generalised Methods of Moments (GMM), the study finds that economic freedom plays
an important role in attracting foreign direct investment. For this reason, the authors argue that
quality domestic institutions are essential for growth.

Empirical literature indicates that there is no consensus on which factors are important for
growth. With mixed findings on what determinants of economic growth in different countries, the
literature on this subject in South Africa is sparse. The current study takes advantage of this
literature gap and include many economic variables, taking into consideration the structural
factors affecting economic growth. Chirwa and Odhlambo (2016) indicates that selecting as
many variables as possible may achieve better results. This study employs the autoregressive
distributed lag (ARDL) model to determine the factors influencing economic growth in South
Africa.

4. Empirical Methodology, Data Description and Sources

This section outlines the research methodology applied. The section presents the description
and sources of data. Further, the econometric model used in this study is outlined.

Sources and Description of Data

This study investigates the determinants of economic growth in South Africa for the period 1994-
2017. The study uses time series annual data. Two models are estimated. The dependant
variable in both model is real GDP (RGDP). The independent variables included in the models
are, portfolio investment (PINV), human capital (HUCT), exports (EXPT), trade openness
(TROP), foreign direct investment (FDIN). In addition to these macroeconomic variables,
institutional variables included in the model are Corruption Perception Index (CPIN) and
Business freedom Index (BSFD). The data for all macroeconomic variables is obtained from
South African Reserve Bank, while data for institutions is obtained from Transparency
International and The Heritage Foundation respectively. Annual data from 1994-2017 is used.
The macroeconomic variables are measured in millions of Rands.
Model Specification

7
The econometric model used in this study is the Autoregressive Distributed lag (ARDL) model.
Although there are econometric models which can be used to determine the long-run
relationships such as Johansen and Engle-Granger cointegration tests, these tests are not
compatible for small sample, due to their power to reject the null hypothesis when it is wrong.
Moreover, to use these cointegration tests, all variables should be integrated of the same order.
For these reasons, the ARDL model is preferred over the Johansen and Engle-Granger
cointegration model as it is well suited for small samples (Ahmad & Wajid 2013). Moreover,
ARDL allows one to work with variables that not integrated of the same order. As long as the
variables are integrated of order 0, I (0) or order 1, I (1). However, the model may disintegrate
if the order of integration surpasses I(1). The following model is estimated:

𝑙𝑛𝑌𝑡 = 𝛼0 + 𝛼1 𝑙𝑛𝑃𝐼𝑁𝑉𝑡 + 𝛼2 𝑙𝑛𝐻𝑈𝐶𝑇𝑡 + 𝛼3 𝑙𝑛𝐸𝑋𝑃𝑇𝑡 + 𝛼4 𝑙𝑛𝑇𝑅𝑂𝑃𝑡 + 𝛼5 𝑙𝑛𝐵𝑆𝐹𝐷𝑡 + 𝛼6 𝑙𝑛𝐶𝐼𝑃𝑁𝑡 +

𝜀𝑡 (1)
𝑙𝑛𝑌𝑡 = 𝛼0 + 𝛼1 𝑙𝑛𝐹𝐷𝐼𝑁𝑡 + 𝛼2 𝑙𝑛𝐻𝑈𝐶𝑇𝑡 + 𝛼3 𝑙𝑛𝐸𝑋𝑃𝑇𝑡 + 𝛼4 𝑙𝑛𝑇𝑅𝑂𝑃𝑡 + 𝛼5 𝑙𝑛𝐵𝑆𝐹𝐷𝑡 + 𝛼6 𝑙𝑛𝐶𝐼𝑃𝑁𝑡 +
𝜀𝑡 (2)

Where 𝑙𝑛Y is the natural log of real GDP. The description and sources of all other variables is
outlined on the section above. 𝜀𝑡 is white noise random errors which is assumed to have 0
mean, constant variance and no serial correlation. All variables are converted to natural
logarithm to linearise the exponential variation in the data.

The ARDL model

The ARDL model is carried out in two steps. First, it is tested if there exists a long-run
relationship between the variables. The F-statistic is used to check the existence of long-run
relationship between the variables. If the calculated F-statistic is higher than the upper bound of
critical values, the null hypothesis that the variables are not cointegrated is rejected, implying
that there exists a long-run between the variable. On the hand, if the calculated F-statistic is
lower than the lower bound, the null hypothesis that the variables are not cointegrated cannot
be rejected, and therefore only short-run relationship between the variables exists (Narayan &
Smyth, 2004). To test for long-run relationship between the variables, the following ARDL
equation is estimated:

𝐾 𝐾 𝐾 𝐾

𝑙𝑛𝑌𝑡 = 𝛼0 + ∑ 𝛼1 𝑙𝑛 𝑌𝑡−𝑗 + ∑ 𝛼2 𝑙𝑛 𝑃𝐼𝑁𝑉𝑡−𝑗 + ∑ 𝛼3 𝑙𝑛 𝐸𝑋𝑃𝑇𝑡−𝑗 + ∑ 𝛼4 𝑙𝑛 𝐻𝑈𝐶𝑇𝑡−𝑗

𝑗=1 𝑗=0 𝑗=0 𝑗=0
𝐾 𝐾 𝐾

+ ∑ 𝛼5 𝑙𝑛 𝑇𝑅𝑂𝑃𝑡−𝑗 + ∑ 𝛼6 𝑙𝑛 𝐶𝑃𝐼𝑁𝑡−𝑗 + ∑ 𝛼7 𝑙𝑛 𝐵𝑆𝐹𝐷𝑡−𝑗 + 𝜀𝑡

𝑗=0 𝑗=0 𝑗=0
8
 (3)
The same will be done for the second model. If the first step confirms that the exists a long-run
relationship between the variables, the second step involves estimation of parameters of the
long-run relationship. This step involves the estimation of short-run dynamic error correction
models (ECM), which is given by:

𝐾 𝐾 𝐾 𝐾

∆𝑙𝑛𝑌𝑡 = + ∑ 𝛼1 ∆𝑙𝑛 𝑌𝑡−𝑗 + ∑ 𝛼2 ∆𝑙𝑛 𝑃𝐼𝑁𝑉𝑡−𝑗 + ∑ 𝛼3 ∆𝑙𝑛 𝐸𝑋𝑃𝑇𝑡−𝑗 + ∑ 𝛼4 ∆𝑙𝑛 𝐶𝑂𝑁𝑆𝑡−𝑗

𝑗=1 𝑗=0 𝑗=0 𝑗=0
𝐾 𝐾 𝐾

+ ∑ 𝛼5 ∆𝑙𝑛 𝑇𝑅𝑂𝑃𝑡−𝑗 + ∑ 𝛼7 ∆𝑙𝑛 𝐹𝐷𝐼𝑁𝑡−𝑗 + ∑ 𝛼8 ∆𝑙𝑛 𝐶𝑃𝐼𝑁𝑡−𝑗

𝑗=0 𝑗=0 𝑗=0
𝐾

+ ∑ 𝛼9 ∆𝑙𝑛 𝑃𝑅𝐼𝑋𝑡−𝑗 + 𝜃𝐸𝐶𝑀𝑡−1 + 𝜀𝑡

𝑗=0

(4)

Once the ECM has been estimated, Pesaran and Pesaran (1997) suggest conducting a
sensitivity analysis, which involves Cumulative Sum (CUSUM) and Cumulative Sum of Squares
(CUSUMSQ) tests, as well as testing if there is no serial correlation.

5. Main Findings and Discussion

Although ARDL does not compel one to test the variables for unit root, it is desirable to conduct
the unit root test. ARDL model can be carried on variables that are integrated of different order,
I(0) and I(1). However, if ARDL is modelled using variables integrated of higher order such as
I(2), the model may disintegrate. To confirm is there are no variables integrated of higher order,
the ADF and PP unit root tests are conducted. Both tests confirm that variables are I(0) and I(1),
Implying that they are suitable for ARDL model.

9
 Table 2: Unit Root Test Results

Augmented Dickey Fuller Test (ADF unit root test) Phillips-Perron Test (PP unit root test)

Variables Level 1st difference Level 1st difference

RGDP -1.667 (0.4332) -2.812(0.0736)* -1.482)(0.5232) -2.870(0.0658)*

EXPT -1.816(0.3633) -4.522(0.0020)*** -2.287(0.1844) -4.770(0.0012)***

TROP -2.337(0.1699) -4.559 (0.0019)*** -2.769(0.1577) -4.877(0.0009)***

FDIN -2.0277(0.2737) -5.1269 (0.0005)*** -2.218(0.2056) -5.1188(0.0005)***

HUCT -9.986(0.2899) -6.564(0.0000)*** -1.814(0.3642) -656(0.0000)***

CPIN -2.142(0.2311) 9.266 (0.0000)*** -1.932(0.3122) -8.519(0.0000)***

BSFD -1.095(0.6986) -3.919(0.0075)*** -1.095(0.6986) -3.919(0.0075)***

PINV -4.427(0.0115)** -3.231 (0.0632)* -2.921(0.1750) -3.35(0.00842)***

Notes: * denote rejection of the null hypothesis at 10% level of significance; **, rejection at 5% level of
significance; ***1% level of significance.
Critical values (CV) are obtained from Mackinnon (1996) one-sided p-value as determined by Eviews 9.5.
The bandwidth is the Newey-West bandwidth.
Rejection Rule: reject H0 if t-statistic is less than critical values.

In this study, two models were estimated. In the first model the variables included are real GDP
as a dependant variable, while the independent variables are included are exports, trade
openness, foreign direct investment, government expenditure on human capital and the two
institutions variable, namely business freedom and corruption perception index. The second
model, with real GDP still the dependant variable, foreign direct investment is replaced with
portfolio investment to determine which investment is robust for economic growth. The summary
of estimation of these 2 variables is provided in tables 3, 4 and 5.

It is important to determine the optimum lag length when using the ARDL model. The optimum
lag length for both models is selected based on the Akaike information Criterion, Schwarz
Information Criterion and Hannan-Quinn Information Criterion. All criterion selected 1 as the
optimal lag length for both models. The results for lag length selection are presented in table 3
below.

Table 3: VAR Lag Order Selection Criterion

Criterion Model 1 Model 2

10
 K=1 K2 K=1 K2

LogL 66.39744 66.62563 64.23 64.42

AIC -5.939744* -5.862563 -5.723270* -5.642127
SC -5.591238* -5.464271 -5.374763* -5.243834
HQ -5.871712* -5.784812 -5.655238* -5.564376

* indicates lag order selected by the different criterion:

AIC: Akaike information criterion

SC: Schwarz information criterion

HQ: Hannan-Quinn information criterion

Once the optimal lag length has been determined, the ARDL equations can be estimated to
check the existence of long-run relationship between the variables. The summary of estimation
for these models are presented in table 4 below, with their respective results of diagnostic tests.
The diagnostic tests confirm that that both models have no serial correlation, heteroscedasticity,
and variables are normally distributed. To check for stability of the models, the Ramsey RESET
(RR) test was performed and confirms that both models are stable.

To check for long run correlation, the bounds test was performed, obtaining the F-statistic to test
the joint hypothesis that all gradient coefficients of lagged variable are equal to zero. The results
indicate that for the first equation, which incorporates FDI, there is a long-run relationship
between the variables, indicated by the calculated F-statistic higher than the I1 critical bound
value. Further, the bounds test suggests that for the second model, there is no long run
relationship between the variables as the value of F-statistic of 1.5 is lower that the I0 bound.
Since the calculated F-statistic value is lower that the I0 bound critical values, we cannot reject
the null hypothesis of no cointegration

Table 4: Diagnostic Tests and Bounds Tests

Diagnostic test Model 1 Model 2

LM serial correlation test 0.56(0.48) 2.04 (0.18)
Normality test 0.43(0.80) 1.7187 (0.42)
RR stability test 0.22 (0.88) 0.008(0.92)
Heteroskedasticity 0.98(0.48) 0.802(0.62)
Bounds Test
Critical Bound Value I0 Bound I1 Bound I0 Bound I1 Bound
1.99 2.94 2.26 3.35
F-statistic 3.066846 1.555027

The next step is to estimate the long-run and short-run equations. The equations are estimated
using the Ordinary Least Square method, with the selected lag length. Table 5 below summarises
the results for long-run equation as well as short-run equation. Model 1 results indicate that
11
exports are negatively related with economic growth; however, the relationship is not significant.
The variable which has positive correlation with economic growth and is statistically significant
is trade openness. This results supports the findings by Shah et al. (2015) who find that
commodity exporting countries exports raw material rather than value added products. Moreover,
using government expenditure on tertiary education as a proxy for human capital, the results
shows that the two variables are negatively correlated, which is not consistent with the
endogenous growth theory. However, the relationship is not significant. Corruption perception
index has positive impact on economic growth, although not significant. This implies that as
correction perception index improves, economic growth is stimulated. Foreign direct investment
has positive and significant relationship with economic growth. The error correction term of the
model has a negative coefficient, as usually expected, however not significant. This implies that,
if there if there is deviation from equilibrium, the economy does not adjust back to equilibrium
within one year. The results for model 2, which excludes FDI, and incorporates portfolio
investment still confirms that exports are negatively associated with economic growth. This
model does not show any sign of long-run relationship between the variables. Portfolio
investments are short-run in nature and may be reversed easily as compared to FDI. The
portfolio investment has negative relationship with economic growth, however not significant.

Table 5: Long-run and Short-run equations estimates Results

Dependant Variable: RGDP

Variable Model 1 Model 2
Constant 0.010737 0.115270
(0.2746) (0.0688)***
EXPT -0.357413 -0.267082
(0.0669)*** (0.1591)
TROP 0.004819 0.002289
(0.1819) (0.4961)
FDIN 0.030170 -
(0.0351)**
HUCT -0.016721 -0.092799
(0.7210) (0.0431)**
BSFD -0.000118 -0.000548
(0.88) (0.55)
CPIN 0.001192 0.001435
(0.2371) (0.1886)
PINV - -0.006641
(0.5578)
ECM(-1) -0.131710 -
(0.2954)

R2 0.528673 0.613271
Notes: *, ** and *** denote 1%, 5% and 10% level of significance, respectively.

Stability results for both models, checked by the Cumulative Sum (CUSUM) and Cumulative
Sum of Squares (CUSUMSQ) are presented in Appendix A. The plot show that CUSUM and
CUSUMQ statistic for both models are within the critical bounds of 5 percent level of significance.

12
Model 2 CUSUMQ results however shows a slight existence of structural break. The models
however, are reliable as confirmed by the stability tests.

6. Conclusion and Policy Recommendations

This study investigated the determinants of economic growth in South Africa for the period 1996-
2019. Using the ARDL and annual time series data, this study suggests that variables that are
correlated with economic growth in the long run are FDI and exports. FDI has a positive
relationship on growth while exports have negative relationship with growth. Other variables that
have long run relationship with economic growth are not statistically significant. The short-run
equation indicates that portfolio does not have any long-run relationship with economic growth.
Portfolio investment has negative impact on economic growth, as opposed to FDI which has
positive impact. The institutional factors are found to have positive relationship with growth,
although not significant.

The results of this study has some policy implications. Since FDI plays a crucial role in promoting
economic growth, it can be recommended that factors supporting FDI are enhanced. On the
other hand, it is crucial exports are diversified to include products with value added.

13
References
Aghion, P. and Howitt, P.1992. A Model of Growth through Creative Destruction. Econometrica,
Vol. 60, 323-351.
Ahmad, K. and Wajid, S. 2013. What matters for Economic Growth in Pakistan: Fiscal Policy or its
Composition? Asian Economic and Financial Review, Vol 3 (2), 196-215.
Barro, R.J. 1990. Government Spending in a Simple Model of Endogenous Growth. The Journal of
Political Economy, Vol. 98 (5), S103-S125.
Barro, R.J. 1999. Determinants of Economic Growth: Implications of the Global Evidence for Chile.
Cuadernos de Economia, Vol 36(107), 443-478.
Bonga-Bonga, L. and Ahiakpor, F. Determinants of economic growth in Sub-Saharan Africa: The
Case of Ghana, MPRA Paper 66923, University Library of Munich, Germany.
Chirwa, T.G. and Odhlambo, N.M. 2016.Macroeconomic Determinants of Economic Growth: A
Review of International Literature, Vol 11 (2), 33-47.
Combarnous and Rougier. (No Date). Institutions, Socioeconomic Models and Development: An
Overview of the literature and a Methodology. Available:
https://pdfs.semanticscholar.org/9bae/e4ca9b65b60e3a5adcd160b0932616ea432a.pdf
(Accessed: 13 August 2019).
Deaton, A. 1999. Commodity Prices and Growth in Africa. Journal of Economic Perspectives,
Vol.13 (3), 23-40.
Dollar, D. 1992. Outward-Oriented Developing Economies Really Do Grow Rapidly: Evidence from
95 LDCs, 1976-1985. Economic Development and Cultural Change, Vol. 40 (3),523-544.
Fedderke, J. 2018. South African Growth Context, Cause and Consequence. The Journal of Helen
Suzman Foundation, 82.
Ferris, J and Gawande, K. 2003. Coordination Failures and Government Policy: Evidence from
Emerging Countries. The Journal of Development Studies, Vol.39, 84-111.
Fischer, S. 1992. Macroeconomic Stability and Growth. Cuadernos de Economia, Vol. 29 (87).
Khemakhem, M. and Abida, A. 2016. Institutions, Foreign Direct Investment, and Economic
Growth in North African Countries. Proceedings of Business and Management Conferences
3405669, International Institute of Social and Economic Sciences. Available:
https://ideas.repec.org/p/sek/ibmpro/3405669.html (Accessed 17 August 2019).

14
Lehne, J., Mo, J. and Plekhanov, A., 2014, What determines the quality of economic institutions:
Cross-country evidence, Working Paper, No. 171, European Bank for Reconstruction and
Development, London.
Lucas, R.E. 1988. On the Mechanics of Economic Development, Journal of Monetary Economics,
Vol. 22, 3-42.
Mansaray, M. 2017. Applied Econometrics and the Determinants of Economic Growth.
International Journal of Academic Research in economics and Management Sciences, Vol. 6(10).
Masoud, N. 2013. Neoclassical Economic Growth Theory: An empirical Approach. Far East
Journal of Psychology and Business, Vol. 11 (2), 10-33.
Mbulawa, S. 2015. Macroeconomic Determinants of Growth in Zimbabwe. Research Journal of
Finance and Accounting, Vol.6 (2).
Mongale, I.P and Monkwe, K. 2015. The Determinants of Growth in the South African Economy:
CVAR Analysis. Journal of Governance and Regulation, Vol. 4 (3).
Most, S.J., and Vann De Berg, H. 1996. Growth in Africa: does the source of investment financing
matter?, Applied Economics, Vol. 28 (11), 1427-1433.
Narayan, P.K. & Smyth, R. 2004. Dead Man Walking: An Empirical Reassessment of the
Dererrent Effect of Capital Punishment Using the bounds Testing Approach to Cointegration.
Available:
https://www.researchgate.net/publication/4812891_Dead_Man_Walking_An_Empirical_Reassess
ment_of_the_Deterrent_Effect_of_Capital_Punishment_Using_the_Bounds_Testing_Approach_to
_Cointegration (Accessed 16 August 2019).
Ocran, M.K. 2019. Economic Development: Facts, Theories and Evidence. In: Economic
Development in the Twenty-first Century. Palgrave Studies in Economic History. Palgrave
Macmillan, Cham.
Romer, P.M.1990. Endogenous Technological Change. Journal of Political Economy.
Vol. 98 (5), pt. 2.
Shah, S.W.A., Ul haq, A., and Farooq, R.M.A. 2015. Agricultural Exports and economic Growth: A
Case of Pakistan. Public Policy and Administration Research, Vol. 5 (8).
Sokolov-Mladenović,S., Cvetanović, S. and Mladenović, S. 2016. R&D Expenditure and Economic
Growth: EU28 Evidence for the Period 2002-2012. Economic Research, Vol.29 (1), 1005-1020.
Statistic South Africa. 2018. Gross Domestic Product Statistical Release, First Quarter 2018,
available: http://www.statssa.gov.za/publications/P0441/P04414thQuarter2018.pdf (Accessed, 5
July 2019).
Statistic South Africa. 2019. Gross Domestic Product Statistical Release, First Quarter 2019,
available:http://www.statssa.gov.za/publications/P0441/P04411stQuarter2019.pdf#targetText=Rea
l%20gross%20domestic%20product%20(measured,the%20fourth%20quarter%20of%202018.&tar
getText=The%20three%20largest%20negative%20contributors,trade%2C%20catering%20and%2
0accommodation%20industries.
Ulku, H. 2004. R&D, Innovation, and Economic Growth: An Empirical Analysis. IMF Working
Papers, WP/04/185

15
Wanjuu, L.Z. and le Roux, P. 2017. Economic Institutions and Economic Growth: Empirical
Evidence from the Economic Community of West African States. South African Journal of
Economic and Management Sciences, Vol.20 (1), a160.
Zouhaier, H. and Kefi. K. 2012. Institutions and Economic Growth. Asian Economic and Financial
Review, Vol. 2 (7), 795-812.

16
 Appendix A.

Figure 1.A. Plot of CUSUM and CUSUMQ, Model 1

12

-4

-8

-12
05 06 07 08 09 10 11 12 13 14 15 16 17

CUSUM 5% Significance

1.6

1.2

0.8

0.4

0.0

-0.4
06 07 08 09 10 11 12 13 14 15 16 17

CUSUM of Squares 5% Significance

17
Figure 1.B. Plot of CUSUM and CUSUMQ, Model 2

12

-4

-8

-12
05 06 07 08 09 10 11 12 13 14 15 16 17

CUSUM 5% Significance

1.6

1.2

0.8

0.4

0.0

-0.4
05 06 07 08 09 10 11 12 13 14 15 16 17

CUSUM of Squares 5% Significance

18